Investigation of the Thermal Mixing in a T-Junction Flow ...€¦ · Investigation of the Thermal Mixing in a T-Junction Flow With Different SRS Approaches ... High-Cycle Thermal

Investigation of the Thermal Mixing in a T-Junction Flow With Different

SRS Approaches

© 2011 ANSYS, Inc. August 9, 20131

M. S. Gritskevich, A. V. GarbarukSt. Petersburg State Polytechnical University & NTS, Russia

Th. Frank, F. R. MenterPBU, ANSYS Germany, Otterfing

Outline

• The OECD/NEA Benchmark on thermal mixing in T-junctions

• The Vattenfall T-junction experiment

• CFD test case description

• Results


• Results

• Summary

OECD/NEA Benchmark on Thermal Mixing in T-junctions


• OECD/NEA Benchmark, 2009-2010; CSNI Report in 2011 • Odemark, Y. et al., 2009. High-Cycle Thermal Fatigue in Mixing Tees: New Large-

Eddy Simulations Validated Against New Data Obtained by PIV in the VattenfallExperiment. In Proceedings of the 17th International Conference on Nuclear Engineering.

• Mahaffy, J., 2010. Synthesis of Results for the T-Junction Benchmark. In CFD4NRS-3 Conference on Experimental Validation of CFD and CMFD Codes to Nuclear Reactor Safety Issues. Washington, DC, USA, p. 3.

Flow Schematic

T=36°°°°, 6 l/s

∅∅∅∅=0.1mT=19°°°°, 9 l/s

∅∅∅∅=0.14m


• Adiabatic walls

• Water mixing at Re = 1.4⋅⋅⋅⋅105

• Mean and transient wall temperatures

in the fatigue zone: thermal striping

Available experimental data


mean U,V,W & RMS: vertical & horizontal central linesat shown cross-sections

Tmean & RMS: x-lines at walls at 0°°°°, 90°°°°, 180°°°°, 270°°°°

Thermal Striping Phenomenon

• Intensive turbulent mixing downstream the T-junction

• Strong temperature fluctuations near the wall, top & side walls in particular


Expected SRS Model Behavior?

Globally or locally unstable, or even stable?

Jet in cross-flow, SAS T-junction, SAS, bounded CD


�� Experiment: very fast mixing, no large stable vortices

Domain and Grid

Main pipe: -3 ≤≤≤≤ x/D ≤≤≤≤ 20, D=0.14m

Branch pipe: z/d ≤≤≤≤ 3.1, d=0.10m

Z

X

Grid: 4.9M elements, hexahedral

Wall y+: ~4÷÷÷÷6, locally up to 12 in the mixing zone


Wall y : ~4÷÷÷÷6, locally up to 12 in the mixing zone

Timestep: 1 ms �� CFL: bulk ~0.5, mixing zone ~1 ÷÷÷÷ 1.5

SST-SAS without Zonal LES

• Central differences (CD)

• Standard bounded CD (BCD)

• Weakly bounded CD (WBCD)

SST-SAS with Zonal LES

Turbulence Models: CFX


µµµµt/µµµµ

SST-SAS with Zonal LES

• Forcing planes ½∅∅∅∅upstream the junction

• WBCD

Instant vorticity

|∇×∇×∇×∇×u|

Turbulence Models: FLUENT

FLUENT: SAS, DDES and Embedded LES

ELES Setup:

• RANS/LES interface: Vortex method, 1000 vortices

• LES zone: Wall-Modeling LES (WMLES)

Advection

schemes:

CD & BCD


CFD Model Setup

Inlet profiles

• Main pipe (T=19°°°°C, Ubulk=0.58 m/s)

o Velocity and turbulence profiles from calculation in periodic pipe

• Branch pipe (T=36°°°°C, Ubulk=0.76 m/s)

o Velocity and turbulence profiles obtained from calculation in pipe flow to match the measured boundary layer thickness (δδδδ~1cm)


Solver setup

• CD - Central difference orBCD - bounded central difference scheme for advection terms

• Standard scheme for pressure interpolation

• Green-Gauss cell based (GGCB) scheme for gradients

• SIMPLEC with 10 iter. per time step for pressure-velocity coupling

• 2nd order Euler scheme for time discretization

ANSYS CFX Results


Calculation of T-Junction in CFX- Influence of Advection Scheme -

The flow was calculated in ANSYS CFX with SAS and different advection schemes available in the code:

• Central differences (CD)

• Standard bounded CD (BCD)

• Weakly bounded CD (WBCD)

CD, Q=5000 BCD, Q=1000 WBCD, Q=5000


Results are strongly affected by the advection scheme.

CD, Q=5000 BCD, Q=1000 WBCD, Q=5000

Calculation of T-Junction in CFX - Influence of Zonal LES, weak BCD -

Without zonal LES,

Q=1000

With zonal LES,

Q=8000

View from

the top


Different mixing

pattern

Calculation of T-Junction in CFX - Influence of Zonal LES, weak BCD -

Wall temperature in the fatigue zoneTop wall line


ANSYS Fluent Results


Isosurfaces of Q-criterion Colored with Temperature by Different SRS Models

Flow results are very sensitive to numerics

• SAS & BCD scheme �� URANS solution

is obtainedSAS & CD �� LES


• For DDES the effect of numericsis smaller than for SAS but still visible on Q-criterion isosurfaces

• For ELES-WMLES there is virtually no effect of the advection scheme on the solution

Thermal Stripping and Thermal Mixing:Mean and RMS Temperature Contours


• The hot water is strongly cooled downstream of the junction and at X/D=4.6 the flow in the pipe has nearly constant temperature

• The thermal striping phenomenon takes place mostly in the upper part of the mixing layer, where high values of temperature fluctuations (about 0.3·∆T) are observed

• Further downstream, the magnitude of these fluctuations decreases and at X/D=4.6 it is as low as 0.1·∆T with a nearly constant distribution across the section

Mean and RMS Velocity Profiles for Different Models with CD Scheme


All models are able to predict the time averaged mean and RMS velocity profiles with good accuracy, when combined with the CD scheme for advection

• Very good agreement between the results and the experimental data

Mean and RMS Velocity Profiles for Different Models with BCD Scheme


The change of the scheme from CD to BCD does not impair the solution for the DDES and ELES-WMLES approaches

SAS model reverts back to URANS mode when used with the BCD scheme

• The lack of the resolved coherent turbulent structures downstream of the junction results in a significant underestimation of resolved RMS velocities

Mean velocity, horiz. line at x/D=1.6

Fluent


CFX

•••• •••• •••• Experiment

Zonal LES,

weak BCD

Mean velocity, vertic. line at x/D=1.6

Fluent


CFX


Zonal LES,

weak BCD

RMS velocity, horiz. line at x/D=1.6

Fluent


CFX


Zonal LES,

weak BCD

RMS velocity, vertic. line at x/D=1.6

Fluent


CFX


Zonal LES,

weak BCD

Mean and RMS Temperature Profiles for Different Models with CD Scheme


Best results are obtained with the use of ELES-WMLES approach, for which almost perfect distributions of the wall temperatures & TRMS are obtained

• For SAS and DDES models, the results of the wall temperature are noticeably less accurate than those obtained with the ELES-WMLES approach

• The results for the RMS temperature indicate that all models predict RMS temperature fluctuations in good agreement with the data

Mean and RMS Temperature Profiles for Different Models with BCD Scheme


Influence of advection scheme is marginal for ELES-WMLES

For SAS with BCD, the thermal mixing is predicted incorrectly

• The wall temperature is significantly underestimated in all considered wall sections• Similar tendencies, but less severe, are observed for DDES with BCD as well

Mean T, top & front lines (0°°°° & 90°°°°)

Fluent


Top FrontCFX


Zonal LES,

weak BCD

• Marginal influence of

advection scheme for ELES

• SAS with BCD returns to

URANS solution

Mean T, bottom & rear lines (180°°°°&270°°°°)

Fluent


Bottom RearCFX


Zonal LES,

weak BCD

RMS T, top & front lines (0°°°° & 90°°°°)

Fluent


Top FrontCFX


Zonal LES,

weak BCD

TRMS and turbulent mixing

correctly predicted with DDES

and ELES-WMLES/ZLES

RMS T, bottom & rear lines (180°°°°&270°°°°)

Fluent


Bottom RearCFX


Zonal LES,

weak BCD

Summary

• OECD/NEA T-junction benchmark successfully investigated with ANSYS Fluent & ANSYS CFX

• All SRS models are able to accurately predict the mean and RMS velocity profiles, when used with low dissipation CD scheme

• Weak local instability can lead to URANS solution with SAS and slightly more dissipative advection schemes (BCD, HiRes)


slightly more dissipative advection schemes (BCD, HiRes)• Can lead to delayed or not sufficient turbulent mixing• DDES model less sensitive to numerical settings

Best SRS approaches:

� Synthetic turbulence methods:Embedded LES in ANSYS FLUENT, Zonal LES in ANSYS CFX

� Less dependent on the applied advection scheme

• Very good agreement with the experimental data for sensitive Tmean and TRMS flow characteristics


Questions ?

Documents

Investigation of the Thermal Mixing in a T-Junction Flow ...€¦ · Investigation of the Thermal Mixing in a T-Junction Flow With Different SRS Approaches ... High-Cycle Thermal